Synthesis founder: AI tutors could let kids learn 20-30x faster — and school should shrink to a couple hours a day

Summary

Chrisman Frank, co-founder of Synthesis, built the company out of the school Elon Musk created at SpaceX for his own children. Musk hired a teacher named Josh Dawn from a private school in Los Angeles, and Frank — who was the first employee at ClassDojo, which grew from zero to a billion-dollar valuation — visited, saw kids working together to solve problems in the form of games, and partnered with Dawn to scale that model. Synthesis launched during Covid, initially putting that collaborative game-based approach online, and has since built an AI math tutor it considers the core of its product.

The learning-time thesis

Frank's central argument is that formal schooling expanded to fill the hours parents needed covered once they entered the workforce — not because eight hours of instruction is necessary. His claim is that AI tutors could let kids learn 20 to 30 times faster, eliminating homework entirely and compressing structured learning to a couple of hours a day. The rest of the day, in his vision, is freed for exploration, creativity, and the kind of hands-on activity that software can't replicate. His own children attend a charter school two days a week where they build bridges and do school plays. Synthesis wants to be the software partner for schools doing that kind of work, not a replacement for in-person experience.

School choice as the policy lever

Frank argues the most useful thing government can do is let public education dollars follow the student. In California, he notes, it costs roughly $20,000 per year to put a child through the public school system — close to the national average. Allowing families to redirect that funding creates competitive pressure: bad schools lose students, innovative models attract them. His analogy is that education, currently structured like the DMV, could function more like a grocery market.

Physical expansion: deliberate restraint

Frank rules out the hybrid software-plus-real-estate model that several well-funded Silicon Valley startups have attempted, calling it extremely hard to execute. If Synthesis ever does open physical locations, the model would be selective academies in major cities — identifying high-potential kids early, similar to how European soccer clubs run youth academies. He frames that as doing for engineering talent what society already does seriously for sports.

Under-sevens, social skills, and where the AI stops

Synthesis doesn't recruit customers under age seven. Frank's reasoning is that the cognitive load of learning to read faces and operate socially is so high in early childhood that screen-based learning competes with brain resources better spent on human interaction. Social fluency acquired before age seven, he argues, is foundational in a way that's hard to recover later — analogous to native language acquisition.

On the bigger question of whether AI will eventually do children's thinking for them, Frank is skeptical that current LLM trajectories lead there. His view is that being smart, knowledgeable, and able to collaborate in groups remains the target, and that collaborative problem-solving is humanity's own form of superintelligence. Synthesis Teams — the second product layer alongside the AI tutor — puts children in small groups to work through complex scenarios requiring argument, decision-making under uncertainty, and course correction. The explicit goal is engineers who can also operate socially, not just technically.

Engagement vs. effectiveness

On the question of whether education needs to compete with TikTok, Frank's position is that school was already bad before short-form video existed, so the problem isn't primarily about attention competition. He points to Duolingo as a cautionary example: highly effective at early retention and gamification, but hardly anyone who uses it actually learns a language. The right product balances engagement and genuine learning outcomes. His framing is that kids are naturally motivated when they feel ideas making them more powerful as thinkers — and that connecting learning to the next grade level is a poor substitute for that intrinsic signal.